Fragmentation selection strategies in linear scaling methods

Abstract

The chemical motivation and the role of molecular fragmentation schemes in various linear scaling quantum chemical computational methods are discussed, with special emphasis on fragmentation based on the properties of molecular electron densities. The special significance of functional groups in fragment selection, the concept and use of delocalized fragments, the “Procrustes Fragmentation” and “Multi-Procrustes Fragmentation” schemes, and the utility of trigonometric weighting in reducing potential errors due to the bias introduced by fragment selection are discussed. The special fragmentation possibilities implied by the Additive Fuzzy Density Fragmentation Principle, and their application in the context of the Adjustable Density Matrix Assembler (ADMA) method are also discussed.